Secretly using AI at work? “We should be more open about it”

Studies show that people are viewing AI more and more as a colleague. For this occasion, Observant asked ChatGPT to generate an image for this article.

Secretly using AI at work? “We should be more open about it”

Hustinx Prize goes to research on AI in the workplace

01-09-2025 · Science

How is AI affecting workplace dynamics? At the Opening of the Academic Year researcher Roman Briker was awarded the Hustinx Prize for his research on this subject.

From helping to write or proofread essays, analysing papers or application letters to giving relationship advice, AI has quickly made its way into virtually every aspect of our lives – at work, at university and at home. How is this changing the way we behave? “The research field is exploding right now”, says Roman Briker, a Moldovan-German assistant professor at the School of Business and Economics (SBE) with a background in psychology. “Most researchers focus on how people use AI in their own work or personal lives. I’m more interested in its effect on groups – how it affects social interactions. We still don’t know much about that.”

An often overlooked aspect, Briker says by way of example, is the influence your colleagues have on your attitude towards using AI tools like ChatGPT in your work. “We see that if a colleague trusts using AI, you’re more likely to trust it too, especially if it’s a colleague you find nice, smart or helpful. But this effect is weaker, or even absent, if you don’t like the person. In other words, our attitude isn’t just shaped by how well AI works, but also by the social context.”

Roman Briker

Another of Briker’s research questions is what happens to workplace hierarchies and dynamics if your boss is an AI. “So, a computer that gives instructions, evaluates performance and provides feedback. Or one step further: a computer that decides who gets hired or fired, or how much you’re paid.” This may sound futuristic or even dystopian, admits Briker, “but we need to start studying these applications now. You have to try to anticipate where things are heading, or you’ll always be one step behind. Researchers are always a step behind anyway – it can take years to go from starting a study to publishing its results – but the lag is especially pronounced in this field, which is developing at breakneck speed.”

Developing a bond

AI is already displaying abilities researchers thought impossible ten years ago, says Briker. “Take creativity. AI models can now write poems and stories that people judge as more original than human work in blind tests.” Another rapidly growing application is AI as a therapist: apps or robots that keep people company through conversation, or provide support and advice on issues like relationships. “That’s actually the most widely used application at the moment. And developers are making these systems increasingly human-like – robots that move naturally, or computer voices that pause, laugh and interrupt you.”

People are beginning to view AI less as a machine and more as a colleague, friend, coach or leader, observes Briker. “And we’re accepting these new roles. More and more people genuinely collaborate with AI and develop relationships with it. That’s a big difference from older technologies. You don’t bond with your Firefox browser.”

Fear

This makes the idea of an AI boss less far-fetched. “There are already companies publicly claiming to be run by AI, though they are exceptions.” Much more common is what Briker calls “shadow AI”: companies using AI for specific purposes, like screening application letters or other HR tasks, but keeping it under wraps. “They fear backlash. People often believe that they themselves use AI the right way, but we tend to believe that others who use it are lazy or overlook AI mistakes.”

The same is true in the workplace. “At least half of all employees keep quiet about using AI in their work or even lie to colleagues about it. Maybe they fear for their jobs, worried their boss might decide a computer could do their work instead.”

More productive

But that isn’t necessarily the case, says Briker. “Ultimately, I see AI as a powerful, augmenting tool that we can work with, not something that replaces us. It can give you new ideas, act as a sparring partner, and take over tedious tasks. Research has repeatedly shown that people who use AI this way are far more productive.”

That’s why Briker believes companies should introduce clearer AI guidelines for their employees. “People need to know what their managers find acceptable and what’s legally permitted. They’re often unsure about that. I’m in favour of more openness. Let employees experiment. The genie is out of the bottle anyway – otherwise, they’ll just use AI in secret.”

Importantly, this would also help people learn how not to use it. “AI isn’t always 100 per cent accurate, and it can be biased, which is how you end up with discriminatory hiring algorithms. But if you use it wisely and carefully, AI can be far more objective than a biased, racist or sexist human boss.”

Robot as tutor

Imagine a tutor you can turn to any time of day, that never tires and knows you inside out. With personalised AI tutors, this future may be closer than you think. But while it could save human tutors hours of work, it also raises challenges, says Briker. “From an ethical perspective, human oversight makes sense. You want to know what students are asking and what issues they’re running into, and you need to make sure the AI is giving sound advice and accurate answers. But if students know a human tutor is monitoring them, they may be less likely to share their thoughts and questions.”

In studies with UM students, including experiments with the talking service robot Temi, Briker and his colleagues found that students indeed were more likely to use these kinds of AI systems when they knew no one was monitoring them. “That’s a dilemma – you can monitor them, but if that discourages most students from using it, what’s the point? The solution is to let students decide what information is shared, while also teaching them how AI works and why human oversight is important.”

Hustinx Science Award

The Hustinx Science Award (formerly Edmond Hustinx Prize for Science), awarded annually at the Opening of the Academic Year by the Hustinx Foundation, aims to highlight the practical relevance of research and underscore Maastricht University’s importance to the Limburg region. The €15,000 cash prize rotates between faculties. This year, the honour fell to SBE.

Image generated by AI (ChatGPT)

Tags: hustinx,ai,oay2025,chatgpt,colleague,ai boss,tutor,trust,research,instagram

Add Response

Click here for our privacy statement.

Since January 2022, Observant only publishes comments of people whose name is known to the editors.